Joint KL yields horizon-free approximation but an information-theoretic lower bound of order Omega(H) for estimation error in autoregressive learning, with matching computationally efficient upper bounds.
hub Canonical reference
Diffusion Models Beat GANs on Image Synthesis
Canonical reference. 82% of citing Pith papers cite this work as background.
65
Pith papers citing it
2,176
external citations · Pith
Background
82% of classified citations
abstract
We show that diffusion models can achieve image sample quality superior to the current state-of-the-art generative models. We achieve this on unconditional image synthesis by finding a better architecture through a series of ablations. For conditional image synthesis, we further improve sample quality with classifier guidance: a simple, compute-efficient method for trading off diversity for fidelity using gradients from a classifier. We achieve an FID of 2.97 on ImageNet 128$\times$128, 4.59 on ImageNet 256$\times$256, and 7.72 on ImageNet 512$\times$512, and we match BigGAN-deep even with as few as 25 forward passes per sample, all while maintaining better coverage of the distribution. Finally, we find that classifier guidance combines well with upsampling diffusion models, further improving FID to 3.94 on ImageNet 256$\times$256 and 3.85 on ImageNet 512$\times$512. We release our code at https://github.com/openai/guided-diffusion
hub tools
citation-role summary
background 14
baseline 2
method 1
citation-polarity summary
representative citing papers
Classifier-free guidance trades off sample quality and diversity in conditional diffusion models by combining scores from jointly trained conditional and unconditional models.
HCMS partitions multi-head attention into chunks and pipelines them across dual CUDA streams to overlap communication and computation, delivering 10-17.5% speedup over Ulysses for 31K-56K token sequences.
Sparse Context achieves 2-4x faster inference in reference-conditioned diffusion models by fine-tuning with random token dropping and applying task-aware selection at inference time, without loss of visual quality.
ECD reformulates compositional diffusion planning as energy minimization over local bridge potentials, adding a boundary reaction term and a Markov score approximation that runs in linear time.
Introduces Forged Calamity benchmark and shows that fine-tuned and zero-shot synthetic image detectors lose substantial accuracy on unseen generators and disaster types.
Flow Reversal Steering steers flow matching generalist policies by reversing suboptimal actions to nearby better modes, enabling improved zero-shot control, quick distillation, and RL bootstrapping in robotic manipulation.
DCIC uses dual constraints on a diffusion decoder to realize adjustable RDP operating points in neural image compression without extra rate cost.
FlowErase-RL applies GRPO to reformulate concept erasure in flow matching models as reward optimization using a dynamic dual-path mechanism for target suppression and non-target preservation.
Margin-calibrated classifier guidance via Sequence Completion Ranking raises multi-step retrosynthesis solve rates from 16.8% to 95.3% on USPTO-190 and unlocks previously unsolvable targets.
Tempered Guided Diffusion uses annealed SMC to produce consistent particle approximations to the posterior for training-free conditional diffusion sampling, outperforming independent guided trajectories in experiments.
MoZoo generates high-fidelity animal videos with fur and muscle dynamics from coarse meshes by extending video diffusion with role-aware RoPE and asymmetric decoupled attention, trained on a new synthetic-to-real dataset.
Early and late denoising steps in masked diffusion LMs are robust to smaller-model replacement, enabling 17% FLOPs reduction with modest generative quality loss.
DiTs use either a two-stage cross-attention circuit or text-token fusion circuit for spatial relations depending on the text encoder, achieving near-perfect in-domain accuracy but differing out-of-domain robustness.
Measurement-based quantum diffusion models are introduced to recover pure and mixed quantum states via weak measurements, quantum score matching, and Petz recovery maps with error bounds, bridging to classical stochastic reversals.
Proposes an advection-diffusion PDE corruption process with stochastic velocity fields and Lattice Boltzmann solver for diffusion models, generalizing prior PDE methods.
A hierarchical prior-decoder model using CLIP latents generates more diverse text-conditional images than direct methods while preserving photorealism and caption fidelity.
Progressive distillation halves sampling steps repeatedly in diffusion models, reaching 4 steps with FID 3.0 on CIFAR-10 from 8192-step samplers.
Latent diffusion models achieve state-of-the-art inpainting and competitive results on unconditional generation, scene synthesis, and super-resolution by performing the diffusion process in the latent space of pretrained autoencoders with cross-attention conditioning, while cutting computational and
A 3.5-billion-parameter diffusion model with classifier-free guidance generates images preferred over DALL-E by human raters and can be fine-tuned for text-guided inpainting.
SDEdit performs guided image synthesis and editing by adding noise to inputs and refining them via denoising with a diffusion model's SDE prior, outperforming GAN methods in human studies without task-specific training.
A narrative survey that catalogs fifty papers on diffusion-based adversarial techniques across text, vision, and vision-language models, proposes a six-class taxonomy of diffusion roles plus a unified five-dimension evaluation framework, and releases a companion catalog.
Diffusion-based per-view harmonization for lighting-consistent object transfer between 3DGS scenes, using heterogeneous training data and final 3D consolidation.
Latent diffusion model parameterization allows MCMC and SMC to outperform latent-space ESMDA in data mismatch and uncertainty reduction for 3D subsurface DA, while model-space ESMDA produces unrealistic posteriors.
citing papers explorer
-
Autoregressive Learning in Joint KL: Sharp Oracle Bounds and Lower Bounds
Joint KL yields horizon-free approximation but an information-theoretic lower bound of order Omega(H) for estimation error in autoregressive learning, with matching computationally efficient upper bounds.
-
Classifier-Free Diffusion Guidance
Classifier-free guidance trades off sample quality and diversity in conditional diffusion models by combining scores from jointly trained conditional and unconditional models.
-
HCMS: Head-Chunked Multi-Stream Pipeline for Communication-Computation Overlap in Long-Sequence Parallel Attention
HCMS partitions multi-head attention into chunks and pipelines them across dual CUDA streams to overlap communication and computation, delivering 10-17.5% speedup over Ulysses for 31K-56K token sequences.
-
Keep The Essentials: Efficient Reference Conditioned Generation via Token Dropping
Sparse Context achieves 2-4x faster inference in reference-conditioned diffusion models by fine-tuning with random token dropping and applying task-aware selection at inference time, without loss of visual quality.
-
Energy-based Compositional Diffusion Planning
ECD reformulates compositional diffusion planning as energy minimization over local bridge potentials, adding a boundary reaction term and a Markov score approximation that runs in linear time.
-
Forged Calamity: Benchmark for Cross-Domain Synthetic Disaster Detection in the Age of Diffusion
Introduces Forged Calamity benchmark and shows that fine-tuned and zero-shot synthetic image detectors lose substantial accuracy on unseen generators and disaster types.
-
Improving Robotic Generalist Policies via Flow Reversal Steering
Flow Reversal Steering steers flow matching generalist policies by reversing suboptimal actions to nearby better modes, enabling improved zero-shot control, quick distillation, and RL bootstrapping in robotic manipulation.
-
Dual-Constrained Diffusion Image Compression for Operational Rate-Distortion-Perception Optimization
DCIC uses dual constraints on a diffusion decoder to realize adjustable RDP operating points in neural image compression without extra rate cost.
-
FlowErase-RL: Rethinking Concept Erasure as Reward Optimization in Flow Matching Models
FlowErase-RL applies GRPO to reformulate concept erasure in flow matching models as reward optimization using a dynamic dual-path mechanism for target suppression and non-target preservation.
-
Margin-calibrated Classifier Guidance for Property-driven Synthesis Planning
Margin-calibrated classifier guidance via Sequence Completion Ranking raises multi-step retrosynthesis solve rates from 16.8% to 95.3% on USPTO-190 and unlocks previously unsolvable targets.
-
Tempered Guided Diffusion
Tempered Guided Diffusion uses annealed SMC to produce consistent particle approximations to the posterior for training-free conditional diffusion sampling, outperforming independent guided trajectories in experiments.
-
MoZoo:Unleashing Video Diffusion power in animal fur and muscle simulation
MoZoo generates high-fidelity animal videos with fur and muscle dynamics from coarse meshes by extending video diffusion with role-aware RoPE and asymmetric decoupled attention, trained on a new synthetic-to-real dataset.
-
Not All Denoising Steps Are Equal: Model Scheduling for Faster Masked Diffusion Language Models
Early and late denoising steps in masked diffusion LMs are robust to smaller-model replacement, enabling 17% FLOPs reduction with modest generative quality loss.
-
Circuit Mechanisms for Spatial Relation Generation in Diffusion Transformers
DiTs use either a two-stage cross-attention circuit or text-token fusion circuit for spatial relations depending on the text encoder, achieving near-perfect in-domain accuracy but differing out-of-domain robustness.
-
Measurement-Based Quantum Diffusion Models
Measurement-based quantum diffusion models are introduced to recover pure and mixed quantum states via weak measurements, quantum score matching, and Petz recovery maps with error bounds, bridging to classical stochastic reversals.
-
Beyond Blur: A Fluid Perspective on Generative Diffusion Models
Proposes an advection-diffusion PDE corruption process with stochastic velocity fields and Lattice Boltzmann solver for diffusion models, generalizing prior PDE methods.
-
Hierarchical Text-Conditional Image Generation with CLIP Latents
A hierarchical prior-decoder model using CLIP latents generates more diverse text-conditional images than direct methods while preserving photorealism and caption fidelity.
-
Progressive Distillation for Fast Sampling of Diffusion Models
Progressive distillation halves sampling steps repeatedly in diffusion models, reaching 4 steps with FID 3.0 on CIFAR-10 from 8192-step samplers.
-
High-Resolution Image Synthesis with Latent Diffusion Models
Latent diffusion models achieve state-of-the-art inpainting and competitive results on unconditional generation, scene synthesis, and super-resolution by performing the diffusion process in the latent space of pretrained autoencoders with cross-attention conditioning, while cutting computational and
-
GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models
A 3.5-billion-parameter diffusion model with classifier-free guidance generates images preferred over DALL-E by human raters and can be fine-tuned for text-guided inpainting.
-
SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations
SDEdit performs guided image synthesis and editing by adding noise to inputs and refining them via denoising with a diffusion model's SDE prior, outperforming GAN methods in human studies without task-specific training.
-
Adversarial Diffusion Across Modalities: A Fusion Survey of Attacks, Defenses, and Evaluation for Text, Vision, and Vision-Language Models
A narrative survey that catalogs fifty papers on diffusion-based adversarial techniques across text, vision, and vision-language models, proposes a six-class taxonomy of diffusion roles plus a unified five-dimension evaluation framework, and releases a companion catalog.
-
Lighting-Consistent Object Transfer Across Radiance Fields
Diffusion-based per-view harmonization for lighting-consistent object transfer between 3DGS scenes, using heterogeneous training data and final 3D consolidation.
-
Data assimilation for subsurface flow using latent diffusion model parameterization: performance of ensemble-Kalman and Monte Carlo techniques
Latent diffusion model parameterization allows MCMC and SMC to outperform latent-space ESMDA in data mismatch and uncertainty reduction for 3D subsurface DA, while model-space ESMDA produces unrealistic posteriors.
-
MergeTok: Unified Continuous and Discrete Visual Tokenization via Token Merging
MergeTok unifies VAE and VQ tokenizers via token merging to impose semantic alignment on continuous latents and stabilize discrete codebook training, achieving lower rFID on ImageNet-256.
-
Continual Learning in Modern Hopfield Networks with an Application to Diffusion Models
Modern Hopfield energy identifies high-energy samples as more prone to intrinsic forgetting in continual learning, with effective energy-based replay validated in diffusion models.
-
UniVL: Unified Vision-Language Embedding for Spatially Grounded Contextual Image Generation
UniVL unifies vision and language into one mask-rendered input processed by an OCR backbone to condition diffusion models for spatially grounded image generation without a standalone text encoder.
-
Global Convergence of Sampling-Based Nonconvex Optimization through Diffusion-Style Smoothing
Recasts sampling-based nonconvex optimization as smoothed gradient descent to obtain non-asymptotic convergence guarantees and introduces the DIDA annealed algorithm that converges to the global optimum.
-
Compositional Video Generation via Inference-Time Guidance
CVG improves compositional faithfulness in frozen text-to-video diffusion models by steering early denoising steps with gradients from a classifier trained on the model's own cross-attention features.
-
A unified perspective on fine-tuning and sampling with diffusion and flow models
A unified framework for exponential tilting in diffusion and flow models that includes bias-variance decompositions showing finite gradient variance for some methods, norm bounds on adjoint ODEs, and adapted losses with new Crooks and Jarzynski identities.
-
DDA-Thinker: Decoupled Dual-Atomic Reinforcement Learning for Reasoning-Driven Image Editing
DDA-Thinker decouples planning from generation and applies dual-atomic RL with checklist-based rewards to boost reasoning in image editing, yielding competitive results on RISE-Bench and KRIS-Bench.
-
Normalizing Flows with Iterative Denoising
iTARFlow augments normalizing flows with diffusion-style iterative denoising during sampling while preserving end-to-end likelihood training, reaching competitive results on ImageNet 64/128/256.
-
Deepfake Detection Generalization with Diffusion Noise
ANL uses diffusion noise prediction and attention to regularize deepfake detectors for better generalization to unseen synthesis methods without added inference cost.
-
VASR: Variance-Aware Systematic Resampling for Reward-Guided Diffusion
VASR separates continuation and residual variance in reward-guided diffusion SMC, using optimal mass allocation and systematic resampling to achieve up to 26% better FID scores and faster runtimes than prior SMC and MCTS methods.
-
Rethinking Forward Processes for Score-Based Nonlinear Data Assimilation in High Dimensions
MASF redesigns the forward diffusion process to align with measurements, yielding a theoretically grounded likelihood score and up to 28.2x speedup on O(10^5)-dimensional Kolmogorov flow under sparse and nonlinear observation operators.
-
Diffusion Models Memorize in Training -- and Generalize in Inference
Diffusion models overfit denoising loss at intermediate noise but generalize in inference as model error smooths the flow field and sampling paths avoid memorized noisy training data.
-
Steerable Vision-Language-Action Policies for Embodied Reasoning and Hierarchical Control
Steerable VLAs trained on rich synthetic commands at subtask, motion, and pixel levels enable VLMs to steer robot behavior more effectively, outperforming prior hierarchical baselines on real-world manipulation and generalization tasks.
-
Navigating the Challenges of AI-Generated Image Detection in the Wild: What Truly Matters?
The ITW-SM dataset and targeted optimization of detector design choices yield a 26.87% average AUC improvement for state-of-the-art AI-generated image detectors under real-world social media conditions.
-
HazeMatching: Dehazing Light Microscopy Images with Guided Conditional Flow Matching
HazeMatching adapts conditional flow matching with hazy-image guidance to dehaze microscopy images while balancing fidelity and realism on synthetic and real data.
-
Characterizing higher-order representations through generative diffusion models explains human decoded neurofeedback performance
NERD uses RL-trained diffusion models on fMRI data to model higher-order uncertainty representations, outperforming controls and linking individual differences to neurofeedback success.
-
Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets
Stable Video Diffusion scales latent video diffusion models via text-to-image pretraining, video pretraining on curated data, and high-quality finetuning to produce competitive text-to-video and image-to-video results while enabling motion LoRA and multi-view 3D applications.
-
Improved Techniques for Training Consistency Models
Improved consistency training techniques achieve FID scores of 2.51 on CIFAR-10 and 3.25 on ImageNet 64x64 in one sampling step, outperforming prior consistency training and distillation methods.
-
SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
SDXL improves upon prior Stable Diffusion versions through a larger UNet backbone, dual text encoders, novel conditioning, and a refinement model, producing higher-fidelity images competitive with black-box state-of-the-art generators.
-
Shap-E: Generating Conditional 3D Implicit Functions
Shap-E encodes 3D assets into implicit function parameters then uses a conditional diffusion model to generate new ones from text, enabling fast multi-representation 3D asset creation.
-
Vector-quantized Image Modeling with Improved VQGAN
Improved ViT-VQGAN enables autoregressive Transformer pretraining on ImageNet tokens to reach IS 175.1 and FID 4.17 for generation plus 73.2% linear-probe accuracy, beating prior iGPT models.
-
Guided Action Flow: Q-Guided Inference for Flow-Matching Vision-Language-Action Policies
Q-guided reverse-time flow sampling with a rollout-trained action-chunk critic improves frozen SmolVLA success on LIBERO, with strong single-task gains but only modest locked held-out transfer.
-
Proposal-Conditioned Latent Diffusion for Closed-Loop Traffic Scenario Generation
Proposal-conditioned latent diffusion generates controllable closed-loop traffic scenarios with improved efficiency and test-time guidance on the Waymo Open Motion Dataset.
-
Multiscale reconstruction of protein conformations from cryo-EM images
A multiscale optimization method using explicit protein backbone geometry reconstructs atomic models from cryo-EM data, showing improved RMSD and TM scores on three simulated datasets.
-
Variational Proximal Policy Optimization
VP2O maps PPO to SVGD in a MoE architecture using functional kernels and expert orthogonalization, claiming +179 ELO on Codeforces and 32% token reduction on AIME for a 33B/4B model.
-
Mapping Whisper Representations to Human ECoG Responses with Interpretable Time-Resolved Neural Encoding
The paper introduces a time-resolved neural encoder combining Whisper embeddings with recurrent temporal modeling and soft attention to predict ECoG responses, finding strongest alignment in intermediate layers and anatomically coherent phoneme organization in electrodes.